Difficulty
The computer opponent AI in Sins of a Solar Empire has adjustable difficulty settings that may be selected at the start of each game, individually for each AI opponent. The available levels are Easy, Normal, Hard, Unfair, Cruel, and Vicious, in increasing order of difficulty. Furthermore, the individual AI players can be set to one of several "playstyles" or "personalities" independent of its difficulty; these will determine what that particular computer player will focus on. The available AI personalities are Aggressor, Fortifier, Researcher, and Economist, focusing mostly on what the names suggest. Both difficulty and personality can also be left to be randomly selected by the game for any or all AI players. Furthermore, the difficulties and personalities chosen need not match across AI players in any way -- although AIs even one level higher will usually defeat any lower-level AIs quite early on if at all close by on the map. This can be counterbalanced by setting several lower-level AIs into a locked team against a single higher-level AI (or a smaller team), but the value to the human player of doing that if neither team includes them is questionable. AI difficulty can be changed mid-game: When loading a saved game, the (human) player is presented with the AI difficulty selection screen again, and changes made here are applied going forward once the game is loaded. This change may be repeated an unlimited number of times in a single match, but always requires saving and reloading. Difficulty Level Overview Basics Caveat: AI difficulty-related information is located in non-moddable portions of the game's code and information released by developers is sparse, if any; therefore, most of the following is unconfirmed information compiled from player tests. Unless indicated otherwise, all data is valid for Rebellion version 1.82; earlier versions of the game (pre-Rebellion in particular) seem to have used different settings for some of the parameters. Given the general difficulty of programming artificial intelligence that would be competitive with a human's generalized intellect at this sort of game, the principal mechanic for adjusting AI difficulty in Sins of a Solar Empire is, as in many strategy games, tweaking the AI's resource income. The base modifiers seem to be as follows: * Easy: 75% * Normal: 100% * Hard: 200% * Unfair: 300% * Cruel: 500% * Vicious: 700% These are multiplicative modifiers relative to the human player (i.e., Normal has the same income as a human would, Hard has twice as much, etc., while Easy gets a one-quarter handicap relative to humans) and apply to the base incomes of all three resources. Given these percentages, it is apparent that high-level AIs have a massive income advantage compared to a human; so much so that at the highest levels their fleet-(re)building capacity is virtually limitless and they can rush a human early on with several times the human's fleet strength. On the other hand, low-level AIs, i.e., Easy and possibly Normal, in addition to their non-boosted economies, also do not use the full range of AI programming present in the game, resulting in poorer decision-making. Some sources suggest Hard is the first level that uses the full AI, while others hold that only Easy uses sub-par AI and the observed difference in gameplay quality (as opposed to quantity) between Normal and Hard is due to Normal being hamstrung by an inefficient economy, with the result of it hardly ever using some of the AI programming's "higher functions". Additional AI advantages Additionally, higher-level AIs seem to benefit from several secondary advantages; the most consistently reported being (in no particular order): * Instant completion of some research topics (list probably hard-coded) * Ability to buy capital ship experience levels beyond the level 4 maximum available to players. New AI capitals up to level 7 have been observed against high-level AIs; Hard, for example, seems to stop at level 5. * Research progresses while required research buildings are queued / under construction; construction with research prerequisites progresses while required research is queued / in progress. (But see Note 1 below.) * Ability to upgrade starbases beyond the 8-upgrade limit applying to humans. Starbases with 10-14 upgrades in place have been reported in testing. (See Note 2 below.) * Diplomatic relationship bonus with other AI players, larger on higher difficulty. This makes it more likely that, in open-team games, harder AIs will gang up on the human, often even across race / faction lines. Note 1: Construction will only spawn once research has been completed and research will only complete once research buildings are finished. However, compared to a human, whose construction and research can be queued in this manner but do not progress, the AI effectively gets the research / build timers running concurrently on prerequisites and dependents. Combined with enough resource income to afford it all, this increases high-level AIs' development speed beyond anything a human can possibly achieve. Note 2: The total number of upgrades starbases can choose from is 17 for the Argonev, 19 for the Orkulus, and 20 for the Transcencia, although out of that total, some are utility upgrades with no direct combat value (2, 3, and 6 for the respective starbase classes). Even so, an extra 2-6 upgrade slots can make a starbase inordinately powerful. Unaffected by difficulty While, as outlined above, the AI gets bonus combinations that amount to an overwhelming advantage for the highest levels, there are many key properties that harder AIs do not ''get bonuses for; for example, ship statistics (hull, shields, armor, antimatter, damage, movement and jump speed, abilities, etc.), planet capacities (logistic, tactical, population), fleet capacity and capital ship crews, upkeep costs, etc. The AI also does not seem to get any free ships or structures of any kind, nor any particular bonuses to colonization and planet development, construction and research speed (excepting the queuing tweak described above), and so on. Likewise, all prices are the same for the AI as they would be for a human player. In general, while the above list of items the AI does get a bonus for might not be exhaustive (due to its experimentally-derived nature and lack of official documentation), players should generally be safe in the assumption that anything not listed is the same for the AI as for a human, regardless of difficulty. AI Personalities, Races, and Factions While not directly intended to be more or less difficult, but rather, to provide variety in the AI's playstyle, the personality chosen for an AI might still have a noticeable impact on the quality of its play. Furthermore, the race (TEC, Advent, Vasari) and faction (Loyalist, Rebel; ''Rebellion ''only) chosen for the AI may interact with its playstyle in various ways, so that some AI personalities do particularly well (or poorly) playing a certain race and / or faction. Follows a brief overview of play-testing information on AI behavior in relation to personality and race / faction combination: Aggressor Prefers direct fleet combat. Intent is to rush ship production and attack other players early on. Expands rapidly; however, less attention is paid to research and building up the economy, which tends to lead to poor late-game performance. Also does not spend much on fortification, making it prone to being steamrolled by a strong opponent. Is at its best playing as the TEC, preferably Rebel. While all AIs generally tend to perform poorly playing as Advent due to the importance of ability synergies (which are mostly beyond the AI's sophistication level) for this race, Aggressor AIs may do better than some other personalities, particularly as Loyalist. Fortifier Essentially a variant on the Aggressor build that pays more attention to area control and empire development. For this reason, tends to be very strong in the late game, when it has a sound economy and also uses starbases extensively. Despite the name, does not tend to "turtle" (fortify a small empire without further expansion); in fact, Fortifiers expand no slower than Aggressors. Tends to be weak in early to mid game due to lower ship construction and more spending on fortification and empire development. Very strong as TEC Loyalist; such a build is likely the best suited to the AI out of all the possible combinations. The Fortifier personality tends to work best for the AI overall; it is suitable for any race / faction combination. Also, it is generally the best personality for Advent AIs (both Loyalist and Rebel), although the Vasari Loyalist and in particular the TEC Loyalist builds are typically stronger. Researcher Focuses on research extensively, to the detriment of expansion speed and fleet strength. Also, its strong bias towards combat researches means it rarely improves competitiveness in the late game, as its economy will be lacking. Usually, Researcher AIs will get overrun by other AI types before their research can amount to much, due to general neglect of the rest of the economy. Researchers are mostly a B-choice for the builds where Fortifiers are strongest, i.e., TEC and Vasari (Loyalist preferred, but Rebel works too); interestingly, TEC Rebels (which the AI is generally weak playing for) actually seem to work better as Researcher than as Fortifier. Conversely, either faction of the Advent does not seem to do very well with a Researcher personality. Economist Despite the name does not actually achieve the strongest economy, as it expands the slowest of any personality (often to the point where it almost doesn't) and neglects combat-related research. Has strong income in early game, but curiously spends most of it on ships (which it commands poorly) and fortifications rather than on empire development. As a result, is usually among the first AIs wiped out (along with the Researchers). The Economist personality does not seem to be very well suited to most faction / race builds; however, it is passable as the TEC Loyalist (which has the strongest economy potential in the game; ironically, the Economist personality needs that in order for its economy not to wither by the late game), and a viable B-choice (after Fortifier and shared with Aggressor) for the Advent Loyalist. Discussion From the player impressions above, it seems that the Fortifier preset comes closest to simulating the behavior of a human player, as it tends to achieve a relatively solid balance of expansion, development, area defense, research, and economy, which ultimately allows it to survive into the late game and develop sufficiently to field large, strong fleets. If Fortifier is the most balanced of the AI personalities, then relative to it as the measurement standard, the Aggressor AI focuses more on fleet sorties and less on development, while the Researcher and Economist AIs mostly do the opposite. Of the latter two, Researcher is biased towards research and Economist nominally towards empire development; however, Economist curiously fails at this, its stated specialization, in later game, making it possibly the least well-balanced of the AI personalities. (Despite that, Economist is not necessarily worse in terms of results than Researcher -- although it is hardly better -- as the latter tends to neglect shipbuilding so badly that it is often the first to be defeated.) As a final note, all of the relative strengths and weaknesses of the various factions apply ''to the AI only; for human players, the game is carefully and, most players seem to agree, quite successfully balanced so that no particular race or race / faction combination is "the best" -- while individual players may certainly find that a particular build suits their playstyle more than any other, it is generally equally possible to play successfully with any of them. That the AI tends to systematically be more successful with some than with others is more due to how well the various builds' strengths correspond with the AI's own strengths, and vice versa, than due to a particular race or race / faction choice being inherently superior. References & See also * As noted in text, most of the data herein was obtained from player testing and is not developer-confirmed. * The primary source for Sections 1 and 2 is this guide / factsheet, which interested readers may also peruse for a more in-depth discussion. * Additional information was collected from various online discussion boards, such as the official game forum, the game forum on Steam, and the game subreddit.